Ayush Garla

Led traffic-sign detection from dataset to deployment, benchmarking YOLOv8 vs. a custom CNN and achieving 99% accuracy for real-time use. Built Python E2E pipelines with Roboflow, AWS Glue, S3, and GCP to curate, augment, and monitor images, cutting ingestion time 40% and speeding training 60%. Fine-tuned GPU inference on GCP for a 64% throughput gain and lower costs. Partnered with mobility/R&D, translating deep-learning into embedded sensors and driving clear updates and process improvements.

Automated ops data workflows in Python/PostgreSQL, cutting manual effort 70%. Built end-to-end ETL on VMs, optimizing SQL to boost processing 55% and speed retrieval 40%. Trained RF/XGBoost regressors for 2k+ employees to predict performance and refine bonuses. Shipped Tableau and SQL dashboards for realtime KPIs, accelerating decisions 35%. Partnered with data scientists, leaders and cross-functional teams to deliver scalable systems, and mentored juniors via reviews and best-practice sessions.